Sunday, April 08, 2018

Book Review: Deep Learning with Python, 1st Edition, by Francois Chollet

There are few books, in recent memory, that I have read that have evoked such a joy and sense of excitment in the process of learning new material.  Imagine the sense, as you read, of a master-teacher guiding you step-by-step.

Francois Chollet's recent contribution of "Deep Learning with Python, 1st Edition", through Manning Publications, is a demonstration of a world-class educator combined with the depth and experience of deep subject matter knowledge.

Chollet's attention to detail, conciseness, and ability to communicate and build the reader's knowledge in successive layers are demonstrated again and again - throughout each chapter.

The structure of the book's chapters should give you an appreciation for the skillful writing needed to cover such a breadth of material in 352 pages.

Part I: Fundamentals of Deep Learning
1 - What is Deep Learning
2 - Before we begin: the mathematical building blocks of neural networks
3 - Getting started wtih neural networks
4 - Fundamentals of machine learning

Part II: Deep Learning in Pratice
5 - Deep learning in computer vision
6 - Deep learning for text and sequences
7 - Advanced deep-learning best practices
8 - Generative deep learning
9 - Conclusions

Appendix A - Installing Keras and its dependencies on Ubuntu
Appendix B - Running Jupyter notebooks on an EC2 GPU instance

This book is such an exceptional value for your money spent - that I would urge any organization to buy copies for your engineers, architects, and data analytics teams today. It is an excellent introductory book on Deep Learning - that will provide a solid foundation for your teams to build upon going forward.

Full Disclosure: Manning provided me a copy to review.

Saturday, March 24, 2018

2018-03-24 Saturday - Crypto Invest Summit West, April 30 - May 2nd, LA

I've submitted a request for a Press Pass to the Crypto Invest Summit West, that will be held April 30 - May 2nd, in LA

Goal: Publish interviews with attendees and speakers, exploring the intersection of challenges and opportunities in emerging technology for crypto investing.

Sunday, October 15, 2017

2017-10-15 Sunday - Rabid Agile, A Gentle Criticism

Caveat: Agile is a goodness, but certain practices/applications of it are not appropriate for every size project

I find the most rabid proponents of Agile (ready-fire-aim for every size project) have no appreciation for scale of when that is appropriate (perfect example of NOT: fixed price contracts with vendors, massive development coordination between dozens of service providers, fixed deadlines - driven by market and regulatory forces)

To an untrained eye, and based purely on that external person's observation, one might conclude that the Rabid Agilist revels in flagrant waste.

Assuming an ∞ budget - they fail to recognize that the business has little tolerance for infinite rework.

2017-10-15 Sunday - Self-Organized Criticality - and Cycles of Corrective Collapse

In my reading this weekend - I happened upon the concept of "Self-Organized Criticality" - and that led me to consider the possible implications for evolutionary constraints on the design of some complex software systems.  The idea of "ongoing cycle of corrective collapse" resonates well with my casual observations of software needing to be refactored / rewritten - to simply make it maintainable and testable.

The point here is that organizations, generally, do not plan for such rework - they wait until the "corrective collapse".

"In 1987 a Danish physicist named Per Bak released a landmark paper introducing the concept of self-organized criticality. Bak observed that complex systems draw stability through an ongoing cycle of corrective collapses that keep the overall system from becoming too over-extended."

I've created this posting as a reminder to come back and revisit this idea in the future - I would like to explore it further in some white papers, when I have more time to write.